You know, the first law of thermodynamics was discovered before special relativity.

OK, we're not interpreting “calories in calories out” that literally, but there are non-literal interpretations of it which still are very good approximation. If you interpret it to mean “it doesn't matter to anything at all whether you eat 2000 calories of proteins or 2000 calories of sugars” it's false, but if you interpret it to mean “the total number of calories you eat is usually way more important than where or when you get them from in determining how much weight you'll gain or lose, provided you're healthy enough and your diet isn't terribly unbalanced”¹ it is essentially true, and the main idea behind The Hacker's Diet by John Walker. ETA: And I don't think the next order of approximation should involve lumping wholemeal rice and high-fructose corn syrup together as “carbs” and extra-virgin olive oil and hard margarine together as “fats”.

1. Or, more fancily, “If you e.g. eat epsilon more calories per day from fat and epsilon fewer calories per day from carbs, or epsilon fewer calories for breakfast and epsilon more calories for dinner, all other things being equal, the change in your medium-term weight-loss rate divided by epsilon will be quite small.”